Brønsted Acid Scaling Relationships Enable Control Over Product Selectivity from O2 Reduction with a Mononuclear Cobalt Porphyrin Catalyst.

The selective reduction of O2, typically with the goal of forming H2O, represents a long-standing challenge in the field of catalysis. Macrocyclic transition-metal complexes, and cobalt porphyrins in particular, have been the focus of extensive study as catalysts for this reaction. Here, we show that the mononuclear Co-tetraarylporphyrin complex, Co(porOMe) (porOMe = meso-tetra(4-methoxyphenyl)porphyrin), catalyzes either 2e-/2H+ or 4e-/4H+ reduction of O2 with high selectivity simply by changing the identity of the Brønsted acid in dimethylformamide (DMF). The thermodynamic potentials for O2 reduction to H2O2 or H2O in DMF are determined and exhibit a Nernstian dependence on the acid pK a, while the CoIII/II redox potential is independent of the acid pK a. The reaction product, H2O or H2O2, is defined by the relationship between the thermodynamic potential for O2 reduction to H2O2 and the CoIII/II redox potential: selective H2O2 formation is observed when the CoIII/II potential is below the O2/H2O2 potential, while H2O formation is observed when the CoIII/II potential is above the O2/H2O2 potential. Mechanistic studies reveal that the reactions generating H2O2 and H2O exhibit different rate laws and catalyst resting states, and these differences are manifested as different slopes in linear free energy correlations between the log(rate) versus pK a and log(rate) versus effective overpotential for the reactions. This work shows how scaling relationships may be used to control product selectivity, and it provides a mechanistic basis for the pursuit of molecular catalysts that achieve low overpotential reduction of O2 to H2O.

Introduction

The reduction of O2 to H2O, often called the oxygen reduction reaction (ORR), is among the most important chemical reactions on the planet due to its role in natural and artificial energy production.[1−4] Extensive efforts have been directed toward understanding how this reaction is catalyzed by enzymes in nature and developing heterogeneous and molecular catalysts that enhance the rate and/or improve the thermodynamic efficiency of the reaction (i.e., lower the required overpotential). The four-electron, four-proton (4e–/4H+) ORR is a complex chemical transformation, and controlling the reaction selectivity presents a significant challenge. Generation of deleterious byproducts, such as superoxide or hydrogen peroxide, limits the energetic efficiency of the reaction and creates other complications, such as degradation of the carbon electrode in fuel cells. Formation of water is thermodynamically more favorable than formation of the partially reduced species. For example, the 4e–/4H+ reduction of O2 to H2O has a standard potential of 1.23 V, while the 2e–/2H+ reduction of O2 to H2O2 has a standard potential of only 0.68 V (eqs and 2).[5] Nonetheless, generation of H2O2 is often kinetically favored because this reaction avoids the relatively high kinetic barrier associated with cleavage of the O–O bond.

Molecular catalysts have played an important role in ORR studies because they are amenable to thorough mechanistic investigation, including systematic structure–activity analyses and characterization of reaction intermediates.[6−10] Macrocyclic cobalt complexes are among the most widely studied classes of molecular ORR catalysts,[11−15] with the first examples reported more than 50 years ago.[16] Efforts to control the reaction selectivity have been an enduring focus of these studies. Mononuclear Co-macrocycles often favor the formation of H2O2 from O2 (eq ).[11−15] Although this reaction has been the focus of renewed attention due to growing interest in electrochemical hydrogen peroxide synthesis,[17−19] historical efforts have been primarily focused on selective 4e–/4H+ reduction of O2 to H2O via strategic catalyst design (Figure ). Prominent examples included cofacial porphyrins (a, b) with two redox-active Co centers available to deliver the four electrons needed for O2 reduction to water;[20−22] 4-pyridyl-substituted porphyrins with appended redox-active units, such as [Ru(NH3)5]2+ (c),[23−25] that could provide additional electrons needed for O2 reduction at a single Co center; and “hangman” porphyrins (d) and related complexes capable of guiding proton delivery to the distal oxygen of a CoIII(OOH) intermediate to promote O–O bond cleavage.[26,27] While some of the tailored catalysts in Figure a achieve high selectivity for the production of H2O (e.g., 99% with catalyst a),[20] most generate mixtures of H2O and H2O2. Alongside these relatively sophisticated catalyst designs, certain mononuclear Co complexes have been identified that achieve high selectivity for H2O (up to 99% with the corrole catalyst e).[28−31]

Figure 1

Summary of representative macrocyclic cobalt ORR catalysts reported previously[11−15] and their selectivities for O2 reduction.

Efforts to control ORR product selectivity with molecular catalysts include diverse catalyst classes, including Fe,[32−34] Cu,[35,36] and Mn[37−39] complexes. New catalyst designs commonly feature multinuclear metal complexes[11,13,20−22,35,36,40] or implement other approaches to control the relative rates of electron[41−43] and/or proton transfer[26,27,34,43,44] as a means to facilitate O–O cleavage and avoid H2O2 generation. Elucidation of the factors that contribute to product selectivity are complicated, however, by the manner in which these catalysts are studied. Molecular catalysts are often immobilized on an electrode prior to analysis, and the immobilization method (e.g., physisorption on edge-plane pyrolytic graphite or incorporation in a conductive “ink” containing Nafion polymer, etc.) can influence the H2O/H2O2 product ratio, even with the same catalyst.[45] These observations suggest that functional groups on the electrode surface or within the supporting matrix (e.g., sulfonic acid groups in Nafion), or intermolecular interactions between immobilized catalysts, influence the product selectivity and hinder interpretation of catalyst structure–selectivity data.

Molecular ORR catalysts may be investigated under homogeneous conditions by using a chemical reductant. This approach facilitates mechanistic studies, and its use in the study of mononuclear macrocyclic Co complexes highlights the intrinsic preference for H2O2 formation with these catalysts (cf. Figure b).[46−52] We recently studied a series of Co complexes bearing pseudo-macrocyclic N2O2-type ligands [e.g., bis(ketiminate), salen derivatives] that led to highly selective H2O2 production in methanol with decamethylferrocene (Fc*) as the reductant and acetic acid as the proton source.[51,52] A parallel study showed that H2O is the sole product when p-hydroquinone was used as a combined source of electrons and protons.[53,54] Mechanistic studies of the latter reaction showed that p-hydroquinone intercepts a CoII(HOOH) intermediate, facilitating O–O cleavage prior to release of H2O2. These results prompted us to consider whether similar H2O-selective O2 reduction could be achieved with independent sources of electrons and protons, similar to that required in the electrochemical ORR.

Here, we show that the cobalt(II) porphyrin complex Co(porOMe) (1) (porOMe = meso tetra(4-methoxyphenyl)porphyrin)[9,15] promotes highly selective formation of either H2O2 or H2O, depending on the identity of the Brønsted acid used in the reaction. The ORR thermodynamic potentials, EO and EO, exhibit Nernstian dependences on the acid pKa,[55] while the CoIII/II potential of the catalyst is independent of the acid pKa. Selective formation of H2O is observed when the CoIII/II potential is above the thermodynamic potential for O2 reduction to H2O2 (EO). Kinetic studies show that the logarithm of the catalytic turnover frequency [log(TOF)] scales linearly with the acid pKa but with different slopes for the formation of H2O and H2O2. These observations are complemented by kinetic, EPR spectroscopic, and voltammetric studies, as well as density functional theory (DFT) calculations, that illuminate the mechanistic basis for the formation of the different products. The ability to use scaling relationships to control product selectivity, as demonstrated herein, has broad implications for molecular electrocatalysis.

Results and Discussion

ORR Rates and H2O2/H2O Product Selectivity with Different Brønsted Acids

The reduction of O2 catalyzed by Co(porOMe) 1 was investigated in DMF with decamethylferrocene (Fc*) as the reductant and a series of different Brønsted acids, employing buffered conditions with an equimolar mixture of each acid and its conjugate base. Acid sources included the following (with their abbreviation and pKa in DMF in parentheses): [(DMF)H][ClO4] (DMF-H+, 1.6);[56]p-toluenesulfonic acid (TsOH, 2.5);[57] 1-propanesulfonic acid (C3H7SO3H, 2.9);[56] 2,6-dihydroxybenzoic acid (2,6-(HO)2BA, 3.6);[58,59] trifluoroacetic acid (TFAH, 4.9);[59,60] oxalic acid, [(CO2H)2, 5.9];[56] dichloroacetic acid (DCAH, 7.5);[58,59] and maleic acid [CH2(CO2H)2, 7.9).[58,59] The ORR rates were monitored by following the growth of the optical absorption band at 780 nm, corresponding to the conversion of Fc* to Fc*+ during the reaction (see Figure S5 in the Supporting Information).[15,51,52] Upon completion of the reaction, the ORR selectivity was established by using iodometric titration and a TiIV(O)SO4 colorimetric assay to quantify the amount of H2O2 present in the reaction mixture, similar to that described previously[49,54,61−63] (see also Section VII in the Supporting Information). The product selectivity is high in all cases (≥93%, Table ), but the product identity changes sharply, from H2O2 to H2O, between the pKa values of 2,6-(HO)2BA (3.6) and TFAH (4.9). Given the high product selectivity, the TOFs reported in Table are calculated for the formation of a single product (H2O2 or H2O) and correspond to the initial rates of H2O2 or H2O formation (mM s–1) normalized to the catalyst concentration (mM).

Table 1

Selectivity of the ORR Catalyzed with 1 under Different Buffered Conditionsa

entry	HA/A–	pKa (HA(DMF))	TOF (s–1)	selectivity H2O:H2O2 (%)
1	DMF-H+/DMF	1.6	7.7 × 10–2	7:93
2	TsOH/NaOTsb	2.5	5.5 × 10–2	6:94
3	C3H7SO3H/C3H7SO3Na	2.9	2.6 × 10–2	1:99
4	2,6-(HO)2BA/2,6-(NaO)2BAc	3.6	2.0 × 10–2	3:97
5	TFAH/NaTFAd	4.9	9.0 × 10–3	95:5
6	oxalic acid/lithium oxalate	5.9	7.1 × 10–3	95:5
7	DCAH/NaDCAe	7.5	5.3 × 10–3	98:2
8	maleic acid/sodium malonate	7.9	4.5 × 10–3	98:2

Reaction conditions: [acid] = [conjugate base] = 10 mM, [1] = 5 μM, [Fc*] = 1 mM, 1 atm O2, room temperature.

TsONa: sodium p-toluenesulfonate monohydrate.

2,6-(NaO)2BA: sodium 2,6-hydroxybenzoate.

NaTFA: sodium trifluoroacetate.

NaDCA: sodium dichloroacetate.

The results of these studies, summarized in Table , reveal a direct influence of the acid strength on the reaction rates and product selectivity. A plot of the logarithm of the TOF versus the pKa of the Brønsted acid reveals linear free energy correlations, with a deviation in slope corresponding to the change in product identity from H2O2 to H2O as the acid pKa increases (Figure ). A similar correlation between acid pKa and ORR rates with a molecular catalyst was described recently by Pegis, Mayer, and co-workers,[57] albeit without a change in product selectivity or slope of the correlation.

Figure 2

Scaling relationship between log(TOF) and acid pKa (black; left axis), which exhibits a different correlation for the selective production of H2O2 or H2O (blue; right axis) catalyzed by 1 (see Table for reaction conditions).

Assessment of ORR Thermodynamics and Kinetics

The pKa of the Brønsted acid will influence the thermodynamic O2 reduction potential (i.e., the driving force for O2 reduction), and the thermodynamic potential for the ORR under nonaqueous conditions was determined as developed[55] and implemented[51,64,65] in several recent studies (see Section II of the Supporting Information for details). Briefly, the reference H+/H2 open-circuit potential (OCP) was determined under each of the buffered conditions (Figure S2),[66] and the thermodynamic potentials for the O2/H2O2 and O2/H2O redox couples in DMF were then estimated by adding 0.68 V (for O2/H2O2) and 1.23 V (for O2/H2O) to the H+/H2 OCP. Corrections associated with the free energy for transferring the H2O2 and H2O products of the ORR from water to DMF are negligible because the solvation free energy of H2O2 and H2O in DMF is similar to that in H2O.[67,68]

Application of this methodology provided the basis for a Pourbaix-like diagram correlating the thermodynamic potentials for H+/H2, O2/H2O2, and O2/H2O with the pKa of the Brønsted acids in organic media (filled circles, Figure ). The potentials exhibit Nernstian trends, with slopes of 59 mV/pKa. Meanwhile, the CoIII/II redox potential of 1 (E1/2(CoIII/II)), determined under the same conditions, is unaffected by the identity of the Brønsted acid (open red circles, Figure ). The latter observation suggests that the conjugate base of the acids does not coordinate to the Co center or otherwise alter the CoIII/II potential under the buffered conditions.

Figure 3

Correlations of pKa(DMF) of acids with H+/H2 and O2/H2O2 redox couples. The reduction potential of O2/H2O2 exhibits a Nernstian dependence on the pKa values of acids, while the E1/2(CoIII/II) of 1 is independent of the acid pKa. See Table for the acid identities and their corresponding pKa values. The O2/H2O2 and O2/H2O redox potentials have been adjusted to account for the non-standard-state background concentrations of H2O and H2O2 used in this study. See Section IV of the Supporting Information for considerations.

For molecular electrocatalysts, an effective overpotential (ηeff) may be defined from the difference between the thermodynamic potential for O2 reduction under the reaction conditions (EORR) and the catalyst redox potential that initiates catalytic turnover, in this case E1/2(CoIII/II) (eq ).[69−71] The ηeff values, together with the catalytic ORR rates (i.e., TOFs) presented in Table , enable analysis of the linear free energy relationships (LFERs) between the log(TOF) and the ηeff values (Figure ).[72]

Figure 4

Linear free energy correlations between log(TOF) and the ηeff for the O2 reduction catalyzed by 1 under the conditions summarized in Table . A variation of this plot, with the ηeff values adjusted to account for the background concentration of H2O or H2O2, is provided in the Supporting Information (see Section VIII and Figure S11).

The LFER plot in Figure shows that the ORR rate increases with increasing driving force (i.e., higher ηeff);[33] however, the plot features an inflection point that correlates with the change in the product identity. H2O2 is formed with stronger acids that contribute to higher ηeff, while H2O is formed with weaker acids that contribute to lower ηeff. The inflection point occurs very close to the thermodynamic potential for the reduction of O2 to H2O2 (ηeff = 0.55 V). Thus, H2O2 is the observed product whenever the CoIII/II potential is below the O2/H2O2 potential. On the other hand, H2O is the observed product when the CoIII/II potential is above the O2/H2O2 potential (i.e., where H2O2 production is thermodynamically unfavorable).

Kinetic, EPR Spectroscopy, and Voltammetry Studies To Probe the ORR Mechanism under Strong and Weak Acid Conditions

The different slopes in the log(TOF) versus ηeff plot in Figure implicate a change in the catalytic mechanism that depends on the strength of the acid present in the reaction mixture. In order to explore this hypothesis, mechanistic studies of the ORR catalyzed by 1 were performed with a representative strong and weak acid, [DMF-H][ClO4] (pKa = 1.6) and DCAH (pKa = 7.5). Initial rate data were collected to establish the catalytic rate law under both conditions (Figure , see also Section IX for full details). With [DMF-H][ClO4] as the acid, the catalytic rate exhibits a first-order dependence on [1], [H+], and [O2], but no dependence on [Fc*] (Figure a and eq ). Analogous data with DCAH as the acid revealed that the catalytic rate exhibits a first-order dependence on [1], [H+], and [Fc*], but no dependence on [O2] (Figure b and eq ).

Figure 5

Kinetic data for the reduction of O2 catalyzed by 1 in the presence of (a) DMF-H+ and (b) DCAH as the acid, obtained by monitoring the initial rates of Fc*+ formation. (a) For 2e–/2H+ O2 reduction to H2O2, the rate law exhibits first-order dependence on [Co], [HClO4], and [O2] but no dependence on [Fc*] (eq ). (b) For 4e–/4H+ O2 reduction to H2O, the rate law exhibits first-order dependence on [Co], [DCAH], and [Fc*] but no dependence on [O2] (eq ). See the Supporting Information for detailed experimental conditions.

Electron paramagnetic resonance (EPR) spectroscopic analysis of 1 clearly showed the formation of a Co-superoxide [CoIII(O2•)] adduct under aerobic conditions (Figure a,b). The CoIII(O2•) species reacts rapidly upon the addition of a strong acid, DMF-H+, resulting in the disappearance of most of the EPR signal (Figure c). This behavior is consistent with an autoxidation pathway to generate CoIII species, as has been reported previously.[73−76] In contrast, no reaction was observed between the CoIII(O2•) species and the weak acid, DCAH (Figure d).

Figure 6

X-band EPR spectra of CoII complex 1 in the absence and presence of O2 and, with the latter, in the presence of added acid. (a) EPR spectra of 1 (1 mM) in N2-saturated DMF. g values = 2.050, 2.015, 2.095; A = 15, 15, 20 G. (b) EPR spectra of 1 (1 mM) in O2-saturated DMF. g values = 2.050, 2.015, 2.095; A = 15, 15, 20 G. (c) EPR spectral evidence for reaction of CoIII(O2•) with HClO4. (d) EPR spectral evidence that CoIII(O2•) does not undergo protonation by DCAH. EPR parameters: microwave frequency = 9.46 GHz, microwave power = 10.4 mW, modulation frequency = 100 kHz, and modulation amplitude = 10 G. Temperature = 110 K. See the Supporting Information for additional details.

Differential pulse voltammetry (DPV) measurements provide complementary insights into the identity of the cobalt species present under the different conditions.[77,78] Complex 1 exhibits an anodic peak at 0.28 V (vs Fc*+/0) in DMF under N2, and it is unaffected by the presence of DMF-H+ or DCAH (Figure S20). When a solution of 1 is exposed to 1 atm O2, a new cathodic peak is observed at lower potential (0.08 V) and is attributed to the CoIII(O2•) species detected by EPR spectroscopy (cf. Figure ). Upon addition of [DMF-H][ClO4], a single cathodic peak at substantially higher potential (0.37 V) is evident (Figure a), consistent with the conversion of CoIII(O2•) into a CoIII species derived from autoxidation of Co(II) in the presence of strong acid (postulated above, Figure c). When DCAH is used rather than [DMF-H][ClO4], the cathodic peak associated with the CoIII(O2•) species shifts to slightly higher potential (0.17 V, Figure b), an effect attributed to hydrogen bonding between the acid and the CoIII(O2•) species, similar to observations made previously with a N2O2-ligated CoIII(O2•) complex.[52]

Figure 7

DPV of 1 under aerobic conditions (1 atm air) in the (a, b) absence and (c, d) presence of Fc* (the latter corresponding to catalytic conditions). Conditions: 25–100 μM 1, 5–10 mM acid, 4–6 mM Fc*, 0.1 M [NBu4][ClO4], 10 mL of DMF. See the Supporting Information for full experimental details.

DPV was also used to probe the nature of the resting-state Co species under catalytic conditions (i.e., 1 in the presence of O2, acid, and Fc*). An anodic peak was observed at 0.33 V with DMF-H+ as the acid source (Figure c). This result, together with the first-order dependence of the rate on [O2] and [H+] and zero-order dependence on [Fc*] (cf. Figure a), is rationalized by a CoII catalyst resting state that undergoes reaction with O2 and DMF-H+ in the turnover-limiting step(s) of the catalytic reaction. In contrast, a cathodic peak was detected at 0.36 V with DCAH as the acid source (Figure d), implicating a CoIII catalyst resting state under the weak acid reaction conditions. The zero-order dependence of the rate on [O2] (cf. Figure b) is consistent with a Co/O2 adduct as the resting state; however, the observed redox potential is considerably higher than that of the CoIII(O2•) species (cf. Figure b). Therefore, we tentatively assign the catalyst resting state to a CoIII(OOH) species, similar to that recently identified for the ORR catalyzed by a N2O2-ligated Co complex.[52] The first-order dependence of the rate on [H+] and [Fc*] supports turnover-limiting electron–proton transfer (EPT) to this species, which could take place in a sequential or concerted process.

Free Energy Profiles for the Co(porOMe)-Catalyzed ORR in the Presence of Weak and Strong Acid

The experimental data and mechanistic proposals presented above and summarized in Table provided the basis for DFT calculations to probe the relative free energies of intermediates under the strong and weak acid conditions. Free energy profiles were computed for the reactions with DMF-H+ and DCAH as the acid with Gaussian 09,[79] using the BP86 exchange-correlation functional[80,81] and the 6-31G**[82] electronic basis set with additional diffuse basis functions on select oxygen atoms. The structures were optimized in the gas phase, followed by the calculation of solvation free energies in DMF using the SMD implicit solvation model.[83]

Table 2

Summary of Experimental Studies of the ORR Mechanism

	strong acid conditions	weak acid conditions
product	H2O2	H2O
catalytic rate law	rate ∝ [Co]1[HA]1[O2]1	rate ∝ [Co]1[HA]1[Fc*]1
proposed resting state	CoII(porOMe)	CoIII(OOH) species

A complication associated with the use of DFT-based results to create free energy diagrams to rationalize all of the experimental observations presented herein is that CoII is the source of electrons used for O2 reduction in the catalytic cycle and provides the basis for the effective overpotential of the reaction (cf. eq ). Thus, the overall thermodynamics of O2 reduction are defined with respect to the CoIII/II potential, and computed overpotentials show relatively good agreement with the experimental data (Table ). On the other hand, the relative free energies of intermediates along the reaction pathway (i.e., those observed experimentally and analyzed by DFT methods) will be influenced by the redox potential of the stoichiometric reductant (in this case, Fc*) that delivers electrons to the Co-based intermediates. The simplified free energy diagrams presented in Figure seek to superimpose both of these considerations: (1) the overall reaction thermodynamics derived from the Co(porOMe) catalyst and DMF-H+ and DCAH as representative strong and weak acids and (2) the relative free energies of relevant reaction intermediates, insights into which were gained from DFT calculations (see Figure S25 in Section XII of the Supporting Information).

Table 3

Experimental and Computed Effective Overpotentials for the ORR Catalyzed by 1 To Produce H2O2 (cf. eq )a

	ηeff(O2/H2O2) experiment (V)	ηeff(O2/H2O2) calculated (V)
strong acid (DMF-H+)	+0.24	+0.12b
weak acid (DCAH)	–0.16	–0.28c

See Section XII in the Supporting Information for computational details.

O2 + 2H(DMF)2+ + 2[CoII]DMF → H2O2 + 2DMF + 2[CoIII]DMF2.

O2 + 4DCAH–DMF + 4[CoII]DMF → 2H2O + 4DCA– + 4[CoIII]DMF2.

Figure 8

Simplified free energy profiles for Co(porOMe)-catalyzed O2 reduction with a (A) strong and (B) weak acid, leading to the formation of H2O2 and H2O, respectively. These qualitative free energy profiles incorporate insights from both experimental and computational data, and they are not intended to convey precise quantitative information for reasons discussed in the text.

The free energy diagrams in Figure and the Pourbaix diagram in Figure show how the formation of H2O2 becomes thermodynamically unfavorable when changing from a strong to a weak acid. The acid strength also influences the free energies of the reaction intermediates by changing the driving force for steps involving proton transfer (PT) and electron–proton transfer (EPT). DFT calculations indicate that the free energies associated with the progression from 1 → 1a → 1b → 1, leading to the formation of H2O2, are all downhill with a strong acid, and the corresponding steps leading to formation of H2O as the final product are even more thermodynamically favorable. The selective formation of H2O2 observed experimentally under strong acid conditions indicates that the reaction product is kinetically controlled and may be rationalized by a high kinetic barrier for O–O cleavage in the formation of water. When using a weak acid, formation of H2O2 is thermodynamically unfavorable, and selective formation of H2O is observed. The required O–O cleavage step could proceed via intermediate 1b or 1c. The former pathway involves a relatively high-energy intermediate (1b), but it would allow the kinetically challenging O–O bond cleavage to occur as late as possible in the mechanism where it would also benefit from the increased driving force relative to EPT-induced O–O cleavage from the CoIII(OOH) species (1c). On the other hand, the latter pathway resembles well-established reactivity involving FeIII–OOH intermediates.[41,84,85] Fundamental studies of other Co complexes[86−90] have characterized relevant intermediates and provide a potential starting point for future studies to explore mechanistic questions concerning the O–O cleavage pathway.

Implications for the ORR with Molecular Catalysts

The correlation between reaction rates and driving force in this system is best understood from the perspective of proton transfer, rather than electron transfer steps, as conveyed by the correlations in Figures –4. Specifically, the change in driving force (i.e., the effective overpotential) arises from the change in pKa of the Brønsted acid, which influences the thermodynamic potential for O2 reduction while not affecting the CoIII/II redox potential (cf. Figure ). Thus, stronger Brønsted acids increase the difference between the ORR thermodynamic potential and the CoIII/II redox potential (i.e., the ηeff; see eq ). The influence of the driving force on the reaction rates is evident from the rate laws for the formation of H2O2 and H2O in eqs and 5, both of which feature a first-order dependence on acid concentration. Stronger acids will have a higher concentration of free H+, as defined by the Ka, and thereby lead to faster rates at higher driving force.

The results presented herein demonstrate the ability to use scaling relationships to predict and control the product selectivity for H2O2 and H2O during the catalytic ORR. The thermodynamics of O2/H2O2 and O2/H2O exhibit a Nernstian dependence on the pKa of the Brønsted acid used in the reaction, while the catalyst CoIII/II potential is unaffected by the acid pKa (cf. Figure ). These different correlations make it possible to access conditions that favor either H2O2 or H2O. With weak acids (high pKa), the CoIII/II potential is higher than the thermodynamic potential for O2 reduction to H2O2 (EO), and the ORR selectively generates H2O. With strong acids (low pKa), the CoIII/II potential is lower than the EO potential, and the reaction selectively generates H2O2. This abrupt switch in product selectivity at the thermodynamic EO potential empirically aligns with the effective overpotential (ηeff) for the ORR, which is defined with respect to the E1/2(CoIII/II) (cf. eq and Figure ).

It is perhaps surprising that this binary dependence of product selectivity on the relationship between the ηeff and EO potential has not been demonstrated previously, but multiple considerations rationalize the lack of precedent. First, many ORR studies with macrocyclic Co catalysts have been conducted under strongly acidic, aqueous conditions with immobilized catalysts (cf. Figure a). The catalytic E1/2 values lie below the EO potential in virtually all of these cases. The cofacial Co-porphyrin complex a in Figure a is a rare exception;[91] however, the highly selective formation of H2O in this case has been attributed to the cofacial bimetallic structure, which is thought to promote O–O cleavage.[11,22] The results presented herein, however, reveal that a binuclear catalyst structure is not required to achieve highly selective formation of H2O. Instead, the reaction simply needs to be conducted under conditions that are thermodynamically unfavorable with respect to the formation of H2O2. Otherwise, the H2O/H2O2 selectively will be dictated by the relative kinetic barriers leading to the two products; however, selective formation of H2O will necessarily feature a large overpotential in such cases.

Another consideration is that many homogeneous ORR studies, including those with complexes h–l in Figure b, have been performed in organic solvents. A reliable methodology for defining the thermodynamic ORR potentials under these conditions was established only recently, however, and the ηeff for most ORR precedents in organic solvent was not determined.[45] On the basis of the results described herein, it is likely that the highly selective formation of H2O2 with catalysts h–l in Figure b arises from catalyst CoIII/II potentials that fall below the EO values.

The results of this study also may be compared to a recent study by Nocera and co-workers.[40] The authors analyzed ORR results with a series of Co- and Fe-based molecular catalysts and concluded that “high ORR selectivities for H2O is a result of large effective overpotentials for the reaction, achieved by the use of strong acids.” This conclusion contrasts the results presented herein, which achieve high selectivity for H2O by using a weak acid, resulting in a sufficiently low effective overpotential that formation of H2O2 is thermodynamically unfavorable. The majority of catalyst systems that exhibit high selectivity for H2O in the study by Nocera and co-workers (and the complementary studies by Mayer and co-workers[33]) operate with very high effective overpotentials (>1.2 V). It is possible that reactions with such a large driving force can access kinetically facile O–O cleavage pathways, either via direct O–O cleavage from M–OOH intermediates, similar to 1a, or via two-electron reduction of M–(HOOH) intermediates, analogous to 1b (cf. Figure ).

Safety Statement

No unexpected or unusually dangerous safety hazards were encountered.

Conclusions

This study demonstrates that selective 4e–/4H+ or 2e–/2H+ reduction of O2 may be accomplished with the same monomeric cobalt porphyrin ORR catalyst, Co(porOMe) 1. The switch in selectivity is achieved by varying the pKa of the Brønsted acid used as the source of protons for the reaction. The acid pKa systematically modulates the thermodynamic ORR potentials for O2 → H2O2 and O2 → H2O in a Nernstian manner, but it does not influence the E1/2(CoIII/II) of 1. These differences in scaling relationships provide the basis for precise control over the product selectivity. The thermodynamically controlled formation of water arises when the CoIII/II potential is above the thermodynamic potential for the production of H2O2. In contrast, selective, kinetically favored formation of H2O2 is observed when the CoIII/II potential lies below the H2O2 thermodynamic potential.

It will be important to extend the results of the present study to other catalyst systems, as the specific Brønsted acid scaling relationships identified here are not expected to apply universally. Changes in the identity and/or number of metal ions in the catalyst, the structure and electronic properties of the ancillary ligand, and specific components of the reaction system (solvent, Brønsted acid, conjugate base, etc.) could lead to changes in the reaction mechanism that will, in turn, influence the magnitude or nature of the scaling relationships. Nonetheless, nearly all proton-coupled redox process should be subject to analogous scaling relationships that could provide opportunities to manipulate product selectivity. The present study highlights the value of understanding the thermodynamic properties of such catalytic reactions as the basis for controlling product selectivity. This principle has important implications beyond O2 reduction, extending to CO2 and N2 reduction and alcohol oxidation in addition to a multitude of other proton-coupled redox processes.